盲源分离技术在水声信号中的应用研究

Research on blind separation technology and its application of underwater acoustic signals

通过研究α稳定分布,提出海洋噪声、舰船辐射噪声等符合低阶α稳定分布,在进行对称α稳定分布且独立的随机源信号进行盲分离处理时,若仍然采用高斯分布将不收敛。本文设计了分数低阶矩阵盲源分离算法,最后通过实验对比本文所设计算法和时延分离算法。实验结果表明:本文所设计的算法对于α稳定分布且有独立的随机信号,具有较好的收敛性,能够有效实现盲源信号分离。通过从复杂的海洋噪声中仍能较好地提取出船舱辐射噪声的实验,说明即使信噪比发生变化,源信号分离后相似系数较分离前仍然有稳定地提高。

By studying αstable distribution,This article put forward ocean noise,ship radiated noise consistent with low- level stable distribution. When deal with symmetric αstable distributions and independent random source signal to blind Separation,if you are still using a Gaussian distribution not to converge. So we design matrix fractional lower order blind source separation algorithm. Finally,compared delay separation algorithm and algorithm designed in this paper by experimental. The results show that：algorithm designed in this paper for stable and independent random signal distribution has better convergence and effectively achieve blind source separation. By changing SNR,extract the radiated noise from the complex ocean noise,explain that similarity coefficient still have a stable increase.